An Approach for Creation of Logistics Management System for Food Banks Based on Reinforcement Learning

Food loss has become a serious problem in recent years, especially in developed countries including Japan. Food welfare organizations called Food Banks which distribute still-edible food to needy people have been active in Japan. Here, we sought to develop a logistics management system for the food banks to improve the efficiency of their food delivery. We propose a method to optimize the food delivery schedule of food banks by means of a reinforcement learning algorithm. In our proposed algorithm, an agent aims to learn the policy for delivering food at the lowest cost. We performed computer simulations to evaluate the validity of the proposed method, and the results demonstrated that the agent could acquire the optimal policy in a small virtual environment in many cases.